Gait Neural Network for Human-Exoskeleton Interaction

The exoskeleton is equipment aiming at bringing more convenience to people's daily life. In order to optimize its interaction with the human, responsiveness is indispensable and gait prediction becomes vital. In this paper, we propose the gait prediction method based on temporal convolutional networks (TCN) that named gait prediction net(GPNet). It consists of the intermediate prediction network and the target prediction network. The novel structure of the algorithm can make full use of the historical information of sensors. Then the performance of the GPNet is evaluated on the public HuGaDB dataset, and comparison results proved the superiority. Finally, the experiment is implemented by the inertial-based wearable motion capture system to show the effectiveness of the proposed method.